One of the major problems in designing optical networks is in the placement of the optical components on the network. Design decisions on the inclusion and placement of components need to accommodate not only the initial traffic on the network, but also future network growth, which is often unpredictable. This issue is especially critical for Metropolitan (or city) networks where cost needs to be minimised and growth is especially unpredictable.
The normal approach to the problem is to include enough elements initially to accommodate currently foreseen traffic growth, and to take a part of the network out of commission where additional upgrade needs to be accommodated. This is difficult to do while minimising impact on existing traffic, and is a major operational expense. Taking part of a network out of commission may involve truck-rolling and splicing a new network element into the existing network as and when required. This involves a heavy operational cost and can mean the network ring or line system is out of commission for hours at a time. A further option is to install an in-line connector into the network where it is thought that a future upgrade may be required and then truck-roll and connect a new element when it is actually needed. This also incurs a high operational cost, though the “down time” involved is less than in the previously mentioned case (tens of minutes instead of hours). Finally, it is possible to install a 2×2 optical switch where the future upgrade may be required, the switch being operated to include the new element in circuit when the need actually arises. While this is fast (<10 ms is normally required for such switching), the reliability is lower and there are higher losses, since two passes of the signal occur and in addition two connectors are involved. Furthermore, most such switches require a source of power and will increase the first-in cost of the network.